Crystalloids vs colloids
Are we poisoning our patients?
updated 09/12
For more information and references see the excellent review articles in the ESICM books ‘Controversies in ICM’ and ‘Patient safety and quality of care in ICM’ both of which I have borrowed heavily from.

6S study NEJM 2012 - final proof that modern 130/0.4% starch used for fluid resuscitation in sepsis increases mortality.

If you want to find out more (including about gelatins and albumin) read on:

Colloid - a substance microscopically evenly dispersed throughout another.
Crystalloid - A substance that in solution can pass through a semipermeable membrane eg sugars and salts completely dissolved in water.

Commonly cited reasons for using colloids are:
  • Faster plasma expansion.
  • Less administered volume.
  • Less pulmonary, and other organ, oedema.
  • Benefits on microcirculation, blood rheology and inflammatory mediators.
It is commonly believed that resuscitation with crystalloids requires 3-4x the volume of colloids and takes longer to achieve desired haemodynamic endpoints.
As well as having a larger molecular size than crystalloids, HES binds to endothelial glycocalyx and induces a sealing effect which reduces fluid flux independently of its action on COP.
It is becoming increasingly evident that fluid overload causes organ oedema and dysfunction, increases complications and worsens outcomes.
There is some evidence that HES may beneficially alter inflammatory processes. HES reduces neutrophil-endothelial interaction and restores macrophage integrity with prevention of increased IL 6 levels.
Hydroxyethyl starch inhibits neutrophil adhesion and transendothelial migration. Shock. 2005 Nov;24(5): 434–9.
Effects of hydroxyethyl starch after trauma-hemorrhagic shock: restoration of macrophage integrity and prevention of increased circulating interleukin-6 levels. Crit Care Med. 1995 May;23(5):806–14.

Faster plasma expansion
Systematic reviews and more recently large prospective trials have shown that resuscitation is equally effective with crystalloids and colloids.
CRYSTMAS (Crit Care 2012). (RCT 196 patients). No significant difference in time to reach haemodynamic stability.
VISEP (537 patients, largest trial in this area, stopped early due to renal failure in HES group) – HES vs CSL – ScvO2 and MAP normalised equally fast.
Study in children with shock showed colloids (HES and dextran) exerted a rapid effect followed by rebound increase in vascular leak a few hours later. Overall time to stabilisation the same.
Comparison of three fluid solutions for resuscitation in dengue shock syndrome. N Engl J Med 2005;353:877–889.
Evidence for faster resuscitation is marginal and does not lead to improved outcomes in critically ill patients (SAFE, VISEP).

Less administered volume
Very large studies (SAFE, VISEP) show considerably smaller ratios of crystalloid to colloid than traditionally thought. Evidence is about 1.5 (1.4 for albumin in SAFE, 1.5 for HES in VISEP)
CRYSTMAS - less HES needed than saline to reach stability
but only 300mls.

These 2 theoretical advantages therefore dwindle when clinical evidence considered.
Why is the theory we learnt not borne out by clinical practice?
Traditional theory based on Starling model of microvascular fluid exchange – passive filtration driven by hydrostatic and oncotic pressure gradients across an endothelium with a fixed permeability.
Newer findings suggest fluid flux controlled by complex active mechanisms meaning COP loses its position as the most important component of fluid flux.

Pulmonary oedema
Trials comparing resuscitation with 0.9% saline, 5% albumin, 4% gelatin and HES in septic and non septic ICU patients show that the type of fluid had no effect on pulmonary permeability and oedema (Anaesth Analg 2006, Crit Care Med 2009)
In severe sepsis, no difference in pulmonary SOFA subscore between CSL and HES (VISEP 2008)
No clinical trial has demonstrated a benefit of colloid over crystalloid in ARDS

Other benefits
Some trials showing improved immune function with HES were fabricated by Boldt.
Other experimental and clinical evidence suggests uptake of HES may result in impairment of immune function.
Effects of colloidal resuscitation fluids on reticuloendothelial function and resistance to infection after hemorrhage. Clin Diagn Lab Immunol 1998;5:543–549. Effect of different resuscitation strategies on neutrophil activation in a swine model of hemorrhagic shock. Resuscitation 2004;60:91–99.
Improvement in micro circulation is a long way from proving an outcome benefit.

So….if they give little or no benefit, do they cause any harm?


Renal impairment

Reabsorption of HES into renal tubular cells causes osmotic nephritic lesions (many studies - see ESICM patient safety). Renal plugging by hyperviscous urine may also be implicated in kidney injury.
There is considerable evidence for renal impairment from HES ranging from AKI in septic patients to chronic toxicity with secondary renal failure in liver transplant patients as long as 10 years after HES administration.
VISEP study showed recipients of HES 10% 200/0.5 had higher risk renal failure and double the days on RRT. Renal impairment correlated with cumulative dose of HES (the more they got, the more nephrotoxic it was). Those who received doses below the manufacturers recommended limit still had a higher risk of renal failure.
Prospective multicentre study in Lancet in 2001 showed HES (6% 200/0.6) caused renal failure in septic patients.
HES shown to cause renal osmotic lesions in transplanted kidneys and renal failure in recipients of brain dead organ donors.
A meta-analysis from 2008 showed receiving HES meant more likely to receive RRT.
Cochrane sytematic review 2010:
Significant increased risk of renal failure overall. RR1.5 (1.2-1.87). Even greater significant risk of renal failure in sepsis group. RR 1.55 (1.22-1.96). Non significant risk of requiring RRT overall. Significant risk of RRT in sepsis group. Many studies had only 1 day follow-up which is not long enough to identify AKI. Included positive trials by Boldt with fabricated evidence in favour of HES. Insufficient data to different molecular weight HES.
Systematic review 2009 of critically ill patients (Open Med):
Significant risk of requiring RRT in HES group in sepsis or septic shock and kidney transplant recipients from BD donors. Lower 95% CIs all >1. 4 trials funded by HES manufacturers. Control solutions were other colloids in many of the trials. Included trials by Boldt.
6S study NEJM 2012 - final proof that even 130/0.4% starch causes renal failure with significantly more patients requiring RRT in starch group compared to Ringers.

What about studies ‘proving’ safety?
Studies quoted as ‘proving’ safety of HES solutions on renal function are flawed – they compare them with other HES solutions or other colloids, use short observation periods and inadequate endpoints for renal dysfunction. With observation periods of 5 days or less and serum creatinine levels as markers of renal dysfunction, neither of the sepsis studies above (VISEP, Lancet) would have revealed the deleterious effects of HES.
An observational multicentre study published in BJA in 2007 suggested HES not associated with increased RRT. However HES recipients at baseline had less exposure to RRT and requirements for RRT in the HES group increased. The cumulative dose of HES was also very low (15ml/kg).
Adverse effects on renal function have been reported with various HES solutions in various surgical settings such as GI, cardiac, renal transplant, max fax and caesarean section.


Meta-analysis in CP bypass patients showed post-op blood loss significantly less with albumin compared with HES.
FDA puts a warning label on hetastarch that it is not to be used in CP bypass surgery because of the risk of bleeding.
In France HES 200/0.62 withdrawn from market after cases of fatal cerebral haemorrhage.
VISEP showed HES recipients had a significantly lower platelet count and received more units of PRCs.

Tissue storage

HES solutions taken up by variety of cells and tissues and can remain there for a long time. Duration of storage depends on type of HES and amount transfused. In healthy volunteers 40% of HES 200/0.5 cleared from plasma was not eliminated renally, but stored elsewhere. Journal für Anästhesie und Intensivbehandlung 1998;3. Quartal:42–46.
Tissue storage can manifest as osmotic nephrotic lesions, lysosomal storage disease with organ failure and intractable pruritis. Other hyperosmotic compounds such as contrast media and immunoglobulins cause similar lesions. HES manufacturers and the GIFTASUP guidelines (now withdrawn) clearly acknowledge that colloids should be combined with crystalloids to avoid hyperoncotic renal dysfunction.
In a post mortem study of 12 ICU patients, HES uptake was detected in kidney, liver, lung, spleen, pancreas, lymph nodes and intestinal mucosa.
HES 200/0.5 has been found stored in the placenta.
Pre-eclamptics treated with HES had a higher caesarean section rate with a trend towards shorter pregnancy and more infants requiring oxygen.
Pruritis is a result of HES uptake in the skin – probably in nerve fibres. Usually refractory to treatment and can persist for up to 2 years. Reported incidence ranges from 30-54%. Incidence and severity increase with dose.

Morbidity and mortality

VISEP. No difference in 28 day mortality, trend (p=0.09) to increased 90 day mortality with HES. This trend a result of a subgroup of patients who had high doses of HES and a mortality of 58% compared to 31% in those with a low dose of HES.
2008 meta-analysis showed a trend towards higher mortality with HES.
2009 meta-analysis (Open Med) showed no mortality difference but noted that the high quality trials had trends to higher mortality. 4 trials were funded by HES manufacturers, many of the studies used other colloids as the control and they included trials by Boldt - the trend towards harm would have been more pronounced if his trials were excluded.
Analysis of 20,000 CABG patients showed lower hospital mortality with albumin compared to non protein colloids.
6S study NEJM 2012 - final proof that modern 130/0.4% starch used for fluid resuscitation in sepsis causes increased death at 90 days. Survival curves separate out at 22 days showing prolonged follow up needed to demonstrate a difference (likely due to delayed harm from tissue accumulation).

2008 study of SAH showed worse neurological outcome at 6 months with HES or gelatin compared to crystalloids. Odds for worse outcome increased linearly with dose of colloid while dose of crystalloid decreased unfavourable outcome.
Effects of fluid therapy following aneurysmal subarachnoid haemorrhage: a prospective clinical study. Br J Neurosurg 2008;22:257–268

In liver transplantation, HES was a risk factor for development of sepsis and mortality. HES shown to cause renal osmotic lesions in transplanted kidneys and renal failure in recipients of brain dead organ donors.

HES 6% 130/0.4 (voluven, volulyte, tetraspan)

Many types of HES solutions with different molecular weights, degree and ratios of substitution. Why so many? In an effort to reduce toxicity – the manufacturers take a poisonous fluid and try to make it less poisonous.
HES 130/0.4 recently registered safe by FDA in US. This was done on the basis of non inferiority studies (control fluids mostly other HES or gelatine solutions), with low cumulative doses (mean 41.9ml/kg) and in low risk patients or volunteers. The mean study period was only 2 days.
No studies (until CRYSTMAS and 6S) on the safety of HES 130/0.4 in severe sepsis or ICU patients with, or at risk of, renal dysfunction.
In 2009, chronic renal failure with osmotic nephrotic lesions and interstitial foam cells was reported with a cumulative dose of 81ml/kg HES 130/0.4 (recommended
daily dose 50ml/kg) in a patient with septic shock.
Pruritis associated with all HES solutions including 130/0.4 (FDA).
CRYSTMAS Crit Care 2012. 6% 130/0.4 starch vs saline in sepsis:
Multicentre RCT. 196 patients
Less HES used to reach haemodynamic stability (mean difference 330 mls)
Time to reach stability - no significant difference
No difference in total amount of fluid over 4 days, LOS, SOFA, AKI.
No difference in mortality, coag, pruritis (not powered for these)
Only average of 4 patients per site.
28 day follow up.
Used saline (more acidosis, reduced renal blood flow and perfusion, coag disturbances and interstitial oedema compared to CSL).
Small sample size, limited volume of starch infused and short follow up time mean safety not demonstrated.
6S NEJM 2012. 6% 130/0.4 starch vs Ringers Acetate in sepsis.
Multicentre R parallel group T. 798 patients
Significant increase in death at 90 days in HES group (primary outcome)
Significant increase in RRT in HES group (secondary outcome).
Severe bleeding and doubling of creatinine more common with HES but not significant.
90 day follow up.
Survival curves show separation of the curves at 22 days suggesting delayed harm due to tissue accumulation - shows long follow up needed to demonstrate harm negating ‘safety’ studies.
Proves that starches increase mortality.


Shorter plasma expansion than HES (gone after 1h)
Anaphylactoid reactions x5 more common with gelatin
Well known to impair haemostasis (although less than HES)
Withdrawn by FDA in 1978 due to prolonged bleeding time.
Shown to reduce platelet function and coagulation by specific decrease in VWF and factor 8
Also evidence that impairs renal function in susceptible patients:
  • Elevates creatinine compared to albumin in CP bypass.
  • Gelatin and HES increased markers of renal impairment after cardiac surgery.
  • Some evidence shows a similar incidence of ARF in ICU patients given gelatin and HES 130/0.4 while sepsis study mentioned earlier showed HES worse.
  • Harmful effects of gelatin on renal function may have been masked in the past by using HES as the comparator.
  • In the SAH study above, both HES and gelatin increased the need for PRCs, elevated inflammatory profiles, increased vasospasm, impaired cerebral autoregulation and worsened 6 month outcome.
Suggested evidence based contra-indications for synthetic colloids:

  • Renal failure and those at risk of it.
  • Sepsis
  • Haemorrhage
  • CP bypass
  • Renal and liver transplant donors and recipients
  • Pregnancy
  • Brain injury - TBI and SAH

i.e. all critically ill patients and those undergoing major surgery.


Due to its negative charge, binds to endothelial surface glycocalyx and reduces its permeability independently of COP. Albumin also binds drugs and inflammatory mediators, is a transporter of active molecules, is a free radical scavenger, inhibits PLT aggregation and maintains colloid oncotic pressure. It is the principal extracellular antioxidant.
Serum albumin concentration inversely related to mortality but this does not mean that increasing albumin reduces mortality.
Some anticoagulant effects in vitro but not in vivo.
Went out of favour after highly controversial meta-analysis suggested increased mortality in critically ill patients.
Proved incorrect by subsequent meta-analyses and randomised controlled trials.
SAFE study enrolled 7000 patients to show no outcome difference between saline and albumin.
Reduces the risk of renal failure in large volume paracentesis compared to other fluids.
Albumin combined with frusemide improves oxygenation with a trend to reduced mortality in ALI.
Subgroup analyses in SAFE study showed a trend (p=0.09) for lower mortality in sepsis with albumin.
A systematic review in 2011 in Crit Care Med showed a
significant reduction in mortality with albumin used for fluid resus in sepsis. Still significant with Boldts studies removed.
Excellent long term safety record.
Subgroup analyses showed a significant increase in mortality with albumin in traumatic brain injury and a trend (p=0.09) for lower mortality in sepsis with albumin.
Observational data from one study suggest hyperoncotic albumin (20%) may be harmful in shock (ICM 2008).
The Improvement in the 2011 systematic review was significant with dilute albumin (4 or 5%) but not with 20%.
Unpublished data from a recent trial with 20% albumin with severe sepsis that did not demonstrate a benefit.
Safe in ICU general patients – it is the only colloid with a large randomised controlled trial to demonstrate its safety.
May be of benefit in sepsis and ALI (but only if 4 or 5%?).
Harmful if TBI.
Need results from more trials to justify its use. 3 ongoing trials at time of this review.

Joachim Boldt

Published dozens of studies ‘proving’ benefit and safety of colloids.
Under criminal investigation for his studies - forged signatures of co-authors, conducted trials without approval, failed to get consent, fabricated entire studies.
Nearly 90 have been withdrawn by journals.
The meta-analyses/systematic reviews of colloids include his studies.


Cheap as chips
No immune hypersensitivity
‘Balanced’ so no acidosis
No coagulopathic effects other than via haemodilution
Completely eliminated
Like all fluids will cause oedema and organ dysfunction if too much given.
It is a resuscitation fluid, not a maintenance fluid.


Trials purporting safety of HES are too small or examine only surrogate outcome measures.
Recent meta-analyses conclude there is no benefit with colloids in critical care or general patients and even in subgroups of trauma, burns or post surgery. In these meta-analyses the 95% CI includes the possibility that synthetic colloids substantially increase mortality. All include ‘beneficial’ trials by Boldt.
There is a wealth of evidence of synthetic colloids causing harm – even 130/0.4 HES.
They are much more expensive than crystalloids.
There is now, finally, convincing data from a large randomised parallel trial (6S - NEJM) that modern starch 130/0.4 increases mortality and causes renal failure.

No meaningful difference in resuscitation times.
Only small difference in volume needed - 1.5:1 not 3-4:1.
No hypersensitivity, no coagulopathy, completely eliminated, balanced, cheap, safe.
No difference in pulmonary oedema.


  • Thousands of patients receive resuscitation fluids every day. Small differences between them could therefore have a large overall effect on outcome.
  • Until large scale trials prove their safety, synthetic colloids should not be used in clinical practice.
  • Synthetic colloids should only be used in the context of clinical trials.
  • If you must use a colloid, use albumin.
  • Starches kill people.